Electronic voting: How well did the new machines work?

Consumer Reports   16 November 2004

Now that the 2004 election is over, voting and technology experts are studying how electronic voting machines performed. For months, critics of the machines had warned of security flaws and the absence of a paper trail for use in audits and recounts. But taking into account the sheer number of votes cast and the massive voter turnout on Nov. 2, the reported problems appeared to be few and far between.

"By and large, the day went quite smoothly," said Doug Chapin, director of Electionline.org, a nonprofit, non-partisan group that conducts research on voting issues.

Reports of election incidents on VoteProtect.org, a Web site designed by groups critical of e-voting, showed on November 3, 2004 that more than 1,000 incidents labeled as "machine problems" had been called in by voters, including people from many e-voting precincts. Given that more than 114 million people voted, however, a minuscule number of machine problems were reported.

In the weeks following the election, other reports of glitches also appeared on Weblogs and in the print media, stirring enough suspicion about fraud to prompt more calls for machines with a voter-verifiable paper trail. Six U.S. Congressmen have asked the Government Accounting Office to investigate the efficacy of voting machines and new technologies used in the 2004 election and to review reports of irregularities in machines of all types.

David Dill, a computer scientist at Stanford University who has raised questions about the security of electronic voting, cautioned that the small number of incidents should not quell concerns about the computerized machines. Some problems may not be detected at all. While nonelectronic ballots can be reviewed to determine whether machines accurately determined voter intent, there is no easy way to run such an audit with e-voting technology.

"We have to keep in mind," he said, "that recording a vote for the wrong candidate is something that doesn't show up in the statistics."

Several reported voting-machine incidents were troubling, said Cindy Cohn, legal director for the Electronic Frontier Foundation, a nonprofit civil rights group focusing on technology issues, in a press teleconference on Tuesday afternoon. (The EFF is also one of the groups affiliated with VoteProtect.org.) Reports ran the gamut from downed machines that exacerbated long lines to potential calibration problems with touch screens. "We had a lot of people who said they would go through and vote for Kerry and when the screen came up it showed them voting for Bush," Ms. Cohen said. She added that voters said they could correct the problem before casting their ballots.

Dan Seligson, the editor of Electionline.org, was in southern Florida on Election Day, in counties using e-voting machines made by Election Systems and Software. These machines had been criticized prior to Nov. 2, 2004 for not having an independent audit trail and no paper record. He said that although voting went well in general, he saw evidence of a "lingering mistrust of these machines," with people commenting as they left the polls that they hoped their vote was actually counted.

Several problems reported on the VoteProtect.org site concerned nonelectronic machines that use punch-card ballots and mechanical levers. In the Bronx borough of New York City, where lever machines are used, 32 callers complained. Some of them said that the levers were jammed or stuck.

Mistrust appeared to extend to paper ballots, too. In California, voters in e-voting precincts had the option of using a paper ballot if they wished. Officials also offered paper ballots if a machine failed. Annalee Newitz, the EFF's policy analyst, said that voter suspicion may have derived from statements that companies making electronic voting machines may have made about problems with paper ballots. The suspicions may also have come from a confusion with provisional ballots, which would not be counted if the voter proved to be unregistered.

Mr. Chapin of electionline.org said he expects that the next step for e-voting is a "more nuanced discussion" of how the technology should perform and what regulations would build trust. He added that e-voting experts are also starting to look at innovations in the machinery, such as touch screens that make a ballot and software programs that allow people to verify their votes.

"We have a sense across the country that the security of e-voting machines is a valid policy concern," he said. "The question is, are paper trails the best solution?"